Everyone seems to agree that we need explainable AI — computer scientists, policy makers, and digital literacy programs. Yet, do we really need explainable AI? What does it mean to posit such a need? Who needs it? Which different alternatives to solving problems with AI does it preclude? My colleagues Suzana Alpsancar, Heike M. Buhl, Ingrid Scharlau and myself discuss these issues in our new paper: Explanation needs and ethical demands: unpacking the instrumental value of XAI.